More Fluke 5440B Woes

[wn1fju]

My Fluke 5440B DC Calibrator has been slowly self-destructing.  Here are the (many) symptoms:

1) Let it run for 4 hours, then do internal calibration.  Immediately after the calibration returns, I set it to output 0.0 volts and the output measures about 12 uV. 
I would have expected it to be much closer to 0 volts.  I mean it supposedly just cal-ed itself!

2) After the four hour run and internal calibration, the unit never really stabilizes and will drift (downwards) about 20 uV over the next couple of hours.

3) At zero volt setting, the unit will occasionally trip to standby with an "Output Limit Fault, Output Under Voltage" error.  Resetting the unit makes it fine again.
The voltage can swing to -1 or -2 volts for a second or two when this happens.

4) After many hours, I can sometimes get random errors, such as "Board Ack Fault, Check Pre-amp Board."  Again, resetting it makes it fine again.

5) Internal analog tests show all pass even right after any misbehavior.

6) After a few hours at 1.0 volts, the unit will suddenly drop to about 0.953 volts (and at 10.0 volt setting, 9.53 volts) and stay there.  I am graphing the output
on a Keithley DM6500 and I can see the drop is instantaneous (to the sample rate of the DM6500).  Resetting the unit brings it immediately back to 1.0 volts,
until it happens again in a while.

I first thought maybe the ovens weren't working properly, but I monitored the various test points on the three ovens, and they seem to be acting properly. 
My next instrumentation will be to simultaneously monitor the 1) reference, 2) DAC and 3) output when I am in condition #6 listed above to see what is
actually changing.

But does anyone have a guess for #6?  What could cause the thing to drop from 1.0 to 0.953 suddenly and then recover immediately upon reset?

[Kleinstein]

The first steps usually are to check the supplies.


For the sudden drop with recovery, if looks like a second more or less stable state. This could be something like an amplifier, regulator or similar part that might oscillate (large signal instability and thus more or less stable oscillation one it started). It could also be a kind of latch up (e.g. classic, shortening a supply) or a hanging OP that runs into phase reversal to get stuck.

The rather large offset is a problem that might be easiest to hunt down, but it could also be something like supply "noise".

[Testtech]

+1 to the power supply check, always the first step.

I would also consider relays and PCB edge connectors to be highly suspect. When these kind of problems appear, I like to tap around on the chassis and boards to see if I can induce a problem.

The "ice cube" relays can be cleaned, as can the PCB edge connectors.

Could also be thermal induced cracked solder joints on anything that gets hot.
My money is on bad connections somewhere.

[wn1fju]

Just to close out the thread....

The previous poster was exactly correct.  Inside the thermal oven on the pre-amp board is a tiny circuit board containing a dual-NPN differential amplifier and a
few passive parts.  This board is connected via 6 enamel wires to a connector on the pre-amp board (the connector is also inside the oven).  Whether it was
the solder joints on the tiny board or the connector on the pre-amp board, I don't really know.  But a re-solder of the wires and a cleaning of the connector
seems to have cured all ills.

This took me a while to find.  I had four separate 6-1/2 digit DMMs monitoring the 1) reference, 2) DAC, 3) pre-amp and 4) output circuitry.  When the thing
suddenly jumped from 1.0 to 0.95 volts, the reference and DAC did not move.  So that led me to the pre-amp board.  A few good raps with a screwdriver
handle on the oven could move the output voltage from 0.95 to 1.0 and back again.

It all makes sense, since any thermally induced cracked solder joints, or marginal connections, are made even worse inside the oven.